CoChat

PhymmBL expanded: confidence scores, custom databases, parallelization and more

Open access
Brady, Arthur, Salzberg, Steven · 2011 · Nature Methods
Cited by 124

The DOI matches its Crossref record.

Abstract

To the Editor: PhymmBL1 is a classification system designed for metagenomics experiments that assigns taxonomic labels to short DNA reads. Since the introduction of PhymmBL in 2009, we made extensive changes and added new features, of which we outline the most important ones here (Supplementary Table 1). We also describe results indicating that PhymmBL effectively classifies samples containing mixed eukaryotic and prokaryotic DNA. PhymmBL combines two components: (i) composition-directed taxonomic predictions from Phymm and (ii) basic local alignment search tool (BLAST)-based homology results2. PhymmBL combines these to label each input sequence with its best guess as to the taxonomy of the source organism. Input sequences as short as 100 base pairs can be phylogenetically classified with PhymmBL more accurately than with any other existing method1 including recently introduced methods (Supplementary Note 1). PhymmBL predicts species, genus, family, order, class and phylum for each read, allowing users to arrange results according to levels of specificity relevant to their research goals. We describe how to configure and operate PhymmBL in a parallelized or grid environment in Supplementary Note 2. PhymmBL’s open-source software runs on all UNIX-like systems, is written in Perl and C++ and can be downloaded free of charge (http://www.cbcb.umd.edu/software/phymmbl/; currently version 3.2). To demonstrate PhymmBL’s ability to classify eukaryotic DNA, we classified 2,278,901 short reads (average, 276 base pairs) from a permafrost-preserved woolly mammoth bone metagenome3. Before classification, we added a variety of genomes to PhymmBL’s local database, representing plants, multicellular animals and protests (Supplementary Table 2). We also built models for the elephant (Loxodonta africana) genome (Elephant Genome Project), expecting that woolly mammoth reads would be labeled as elephant. Our goal was to examine whether PhymmBL could identify eukaryotic DNA as accurately as it had identified bacterial DNA. The most abundant label (59.7%) was indeed elephant (Fig. 1). The next three most abundant predictions were Flavobacterium johnsoniae (2.4%), Polaromonas naphthalenivorans (1.6%) and Polaromonas sp. JS666 (1.0%), all three of which are known Arctic bacteria. These likely represent modern bacterial species present on the mammoth bone. PhymmBL therefore effectively separated eukaryotic from prokaryotic reads and accurately predicted the closest relative of the particular eukaryote that was sequenced, despite the presence of other potentially competing eukaryotic genomes in the local database. Details of the computational resources used by PhymmBL are available in Supplementary Note 3. Figure 1 Screenshot of predicted genus-level taxonomic distribution for the mammoth metagenome, focused on genus Loxodonta. PhymmBL output now includes scores reflecting the software’s confidence that its predictions are correct. A confidence score between 0 and 1 appears with each clade-level prediction for each input read. Polynomial functions estimate accuracy, mapping read length and raw score to these normalized confidence scores. For notes on use and algorithmic details of the confidence score computation, see Supplementary Note 4. Users now can add an arbitrary amount of custom genomic data (as single genomes or in batches) to their PhymmBL database, eliminating exclusive reliance on RefSeq bacterial and archaeal genomes. The default classification database can be augmented with private (or even synthetic) genomes or with additional publicly available genomic data. Eukaryotic and viral sequences can also be added, expanding PhymmBL’s classification mandate beyond prokaryotes. For viewing PhymmBL output (Fig. 1), the freely available Krona package (http://krona.sourceforge.net/) provides an interactive viewer (B. Ondov, N. Bergman and A. Phillippy; personal communication). Krona offers an intuitive HTML5 interface that helps users explore predicted taxonomies for metagenomic read sets. Interacting with a dynamic radial visualization (Fig. 1) that changes to highlight designated areas of interest, users can view predictions for their sample population at different levels of phylogenetic granularity, with clade metadata and population statistics displayed alongside.

Overview

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1038/nmeth0511-367

Cites

Statistics

Verification

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.

A few free checks a day · papers already verified open instantly · results are public

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Citations

Cited by

Metadata

Authors
Brady, A., & Salzberg, S.
Venue
Nature Methods
Published
2011-04-28
Volume
8
Issue
5
Pages
367-367
ISSN
  • 1548-7091 (ISSN-L)
  • 1548-7105
OpenAlex
W2040486579
Cited by
124
References
0
FWCI
6.00
Type
journal article
Language
en
Open access
bronze

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.